105
We conducted the 19th Consumer Pyramids Research Seminar on 5th May 2022 at 6.30 PM IST.
Event Summary:
Sutirtha Sinha Roy and Roy Van Der Weide from World Bank presented their paper on estimates of poverty in India. Their paper finds significance in the context of a large gap in data on poverty and inequality in India since 2011.
The authors stated two main objectives of the paper. The first objective was to provide a comprehensive assessment of CPHS. CPHS data was then re-weighted based on this assessment. The second objective was to generate estimates of poverty and inequality since 2011, using the re-weighted CPHS data.
A max-entropy approach was used to re-weight CPHS data using variables of demographics, asset ownership, and labour indicators observed in NFHS and PLFS data. Adjustments were made to the weights using benchmark surveys for state and rural/urban population levels, to attain country-level representation. The re-weighted CPHS data was used in two approaches to study the evolution of poverty in India. The first approach consisted of imputing NSS-type consumption in CPHS. The second approach utilized CPHS consumption data completely. The authors found a reduction in extreme poverty of 12.3 percentage points between 2011 and 2019. Additionally, they observed greater poverty reduction in rural areas. The results were validated with trends observed in Night Lights data, Nielsen retail surveys, IHDS, and banking data from Chodorow et al (2019).
Anup Malani, serving as discussant, provided a simplified yet comprehensive outline of some of the weighting and estimation procedures undertaken by the authors. Malani proposed matching CPHS 2017 and NSS 2017 data as an alternative approach for re-weighting. He also discussed whether income would be a better measure than consumption to estimate poverty. It was noted that poverty estimates from raw CPHS data were quite comparable to the re-weighted estimates. Malani remarked on the importance of investigating errors in inflation and PPP (purchasing power parity). Exploiting the longitudinal nature of CPHS data was also suggested.
The event was attended by over 100 participants, who contributed several questions and comments. Anmol Somanchi, whose critique of CPHS was addressed in the presentation, was in attendance. He posed a question on how well the max-entropy re-weighting procedure achieved representativeness. The authors pointed to technical explanations for this and several other questions, in the appendix of the presentation. They have also provided the link to the weights generated in the paper, in the presentation slides.